Recent advances in coding techniques allow for transporting the data of a broadcast/multicast service onto multiple streams, e.g. alternative (simulcast) or optional (layered multicast). Such approaches have attracted attention of the Internet community for enabling coarse-grained quality adaptation in multicast communication and several works have built on these, as for example DiffServ (Differentiated Services—see Blake et al., “An Architecture for Differentiated Services”, RFC 2475, 1998, all RFCs and Internet Drafts by the IETF are available at http://www.ieff.org), RSVP (see Braden et al., “Resource ReSerVation Protocol (RSVP)—Version 1 Message Processing Rules”, RFC 2209, 1997), or NSIS (see Hancock, “Next Steps in Signaling: Framework”, Internet Draft (draft-ietf-nsis-fw-05.txt), 2003). However, the architecture of 3 G communication networks, e.g. like that of 3GPP networks, differs from that of the Internet and thus demands different or additional solutions.
The increasing diffusion of bandwidth-intensive multimedia applications to heterogeneous groups of users led to intensive research in the area of multicast rate and congestion control in the Internet. Since the pioneer work of McCanne et al. (see McCanne et al., “Receiver-driven Layered Multicast”, Proceedings of ACM SIGCOMM '96, p. 117 to 130, 1996), multi-rate multicast has been considered as a very promising approach for rate adaptation in streaming scenarios. Techniques have been proposed for transmitting the same content using multiple multicast groups mapping onto different quality levels, based on a cumulative layered data organization (hierarchically encoded) or on stream replication (independent and alternative streams). Moreover, a combination of both approaches is also possible. For example, a session of a single audio stream and several alternative video streams encoded with a standard coding scheme at different data rates or robust to different loss rates.
Generally, the Internet Multicast Model provides basic mechanisms for distributing data with different QoS parameters to subsets of the multicast distribution trees. The hosts, which communicate with the multicast routers using the Internet Group Management Protocol (IGMP—see Fenner, “Internet Group Management Protocol, Version 2”, RFC 2236, available at http://www.ieff.org), can in principle actively adapt the QoS in a sub tree by joining/leaving multicast groups.
However, not all communication networks, e.g. mobile communications networks, follow Internet's end-to-end paradigm. In this regard, compliance to the end-to-end principle means that end hosts are responsible for adaptation to network conditions, relying exclusively on implicit network signaling, i.e., packet drops and delay variations.
Mobile communications networks, on the other hand, usually follow a network-centric approach for QoS provision, resulting in a different Broadcast/Multicast Service Model. Subscribed users are allowed to express their interest on a multicast session by IGMP or similar signaling to dedicated network nodes. The data distribution tree along which service data are provided, however, is build autonomously and modified by the network when necessary, e.g. in response to handover. This approach is advantageous in particular since the radio network controller has the knowledge of available resources (e.g. by providing resource control functionality), and it allows end users to be provided with a more or less seamless service.
Network-centric approaches for providing heterogeneous communication services in the Internet have also been developed. A well-known way to place enhanced functionality within the network is the establishment of transport-level or application-level gateways, or the introduction of active network nodes, as presented in Amir et al. “An application level video gateway”, Proceeding of ACM Multimedia '95, San Francisco, Calif., USA, November 1995 or in Metzler et al., “AMnet: Heterogeneous Multicast Services based on Active Networking”, Proceedings of the 2nd Workshop on Open Architectures and Network Programming (OPENARCH99), New York, N.Y., USA, March 1999, respectively.
While the former approach imposes significant overhead due to transcoding operations, the latter approach provides a framework that would have to be adapted in each case to provide network-specific functionalities and mechanisms.
The first concept for a heterogeneous QoS in the MBMS Architecture was proposed as Option G in the 3GPP TR 23.846: “Multimedia Broadcast/Multicast (MBMS); Architecture and functional description (Release 6)”, V6.1.0, December 2002, available at http://www.3gpp.org. It defines a MBMS Bearer Service that may include multiple streams (optional or alternative), each mapping to a single RTP instance. Each stream is transported over a unique tunnel between GGSN (Gateway GPRS Support Node) and RNC (Radio Network Controller), which is maintained throughout the duration of a MBMS service.
Thereby, it is in principle possible for a RNC to choose a stream of a MBMS service at session start as well as changing/adding streams during the session. However, in order to allow for this functionality, appropriate mechanisms have to be implemented in the radio access network (RAN). A necessary prerequisite is the communication and management of stream states and relations, which allows a RAN node to choose the (set of) appropriate streams according to the current conditions of a cell or downstream nodes
The 3GPP Multimedia Broadcast/Multicast Service (MBMS) Architecture currently only supports a very simple QoS model. It basically provides a non-scalable and non-adaptive service, where either all branches of a MBMS distribution tree are established with the same QoS or the whole service is cancelled. There is no negotiation of QoS values between network nodes, which implies that some of the branches may not be established if QoS requirements cannot be met by the according network nodes. This is relevant both at the beginning of a session or during a session, e.g. at each handover, when a new branch of the distribution tree has to be created/torn down.
On the other hand, mobile terminals are quite heterogeneous with respect to their provided capabilities, i.e., processing power, display size, etc. The current MBMS architecture cannot cope with this heterogeneity or it can by subjecting all terminals (those with better and worse capacity) to a worst case scenario, where all adapt to the lowest quality.